[mrtg] Preventing spikes from "wrapping" counters?

[Steve Shipway]

The best way to deal with this is to use MaxBytes and AbsMax to indicate a sensible maxima for these values, so that in the event of a restart, MRTG will think the counter has reset and calculate a huge value, but will reject the value as it is >AbsMax.

Alternatively, if you use routers2 as the frontend, there is a 'rescale' button in the web interface that will scale the graph relative to the average rather than relative to the peaks.  This will not correct the problem with inaccurate Max values, though.

Steve

________________________________
From: mrtg-bounces at lists.oetiker.ch [mailto:mrtg-bounces at lists.oetiker.ch] On Behalf Of Travis Veazey
...
My problem is that any time Shorewall is restarted (e.g. to load configuration changes), the counters all get reset to 0 and mrtg appears to assume that this means it has wrapped around, resulting in huge spikes on our graphs. This makes them near impossible to read (to say nothing of the false "maximum" and traffic totals now reported). Is there any way to get mrtg to not make this assumption that a suddenly-low value means a wrapped counter?

Strangely, this problem does not exist on an otherwise-identical 32-bit system.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.oetiker.ch/pipermail/mrtg/attachments/20100216/ad2dedf8/attachment.htm